Grad students are hopefully semi autonomous and don’t need close supervision.
Another thought, I think ai enabled science is something of a panacea for the publication crisis of too many bullshit boring papers. That’s because no longer is that effort time consuming enough to suggest you need to write a paper. You could do it in two weeks. So now reviewers could just reject work for it doing enough on the assumption a coding agent could have done the work in a couple weeks. I have rejected grants I have reviewed recently on this principle. Why not papers?
Another thought, I think ai enabled science is something of a panacea for the publication crisis of too many bullshit boring papers. That’s because no longer is that effort time consuming enough to suggest you need to write a paper. You could do it in two weeks. So now reviewers could just reject work for it doing enough on the assumption a coding agent could have done the work in a couple weeks. I have rejected grants I have reviewed recently on this principle. Why not papers?